I'm new to python neural networks. I used ml.net in the past. There was easy to take accuracy of each predicted example f.e -> my model recognized bike on photo with acc = 60%.
Now I need to make similar thing in python. I'm using keras and tensorflow.
This is how I make my prediction:
predict = model.predict_classes(data)
It returns predicted class but how to get accuracy of this prediction?
Maybe you are looking for the evaluate function in Keras: https://keras.io/models/model/#evaluate
Some additional specific metric functions: https://keras.io/metrics/#usage-of-metrics
Also, there are some good examples to get you started in case you run into issues: https://keras.io/getting-started/sequential-model-guide/
Hope this helps.
I think what you're looking for is the model.predict(<np.array>)
If your task is classification and you are doing multi-class classification, remember to use softmax at the end to get the correct probabilities.
I think you are looking for probability of each class if it's true so you can use: "predict_proba" for example if your feature victor is "data" in sklearn it would be:
pro=model.predict_proba(data)
Which pro is an array contain probabilities of each class. In your case for example 60% bike, 30 % car and so on.